Globally, approximately 15 million babies are born preterm each year and 1.1 million deaths are due to preterm birth (PTB), defined as delivery of an infant before 37 post-menstrual weeks. Because mortality and morbidity rates are dependent upon gestational age, the very preterm neonate (<32 weeks gestation) is at the highest risk of developing complications that can result in death or significant life-long disability. Among the most significant and common of the major neonatal morbidities are intraventricular hemorrhage (IVH), bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), sepsis, patent ductus arteriosus (PDA) and retinopathy of prematurity (ROP). While measures of neonatal illness severity have been successful in predicting the risk for mortality in very preterm neonates, our ability to identify newborns likely to develop significant morbidity remains limited. Neonatal illness severity indices have a variety of important clinical and research applications including risk stratification, family counseling, external benchmarking for inter-hospital performance comparisons, and determining individual treatments for infants with a specific risk profile. Scoring systems are needed that not only predict mortality but also morbidity in the very preterm neonate. Our team has shown that metabolic status at the time of routine newborn screening is a novel predictor of neonatal morbidity and mortality in preterm newborns. Further work is needed to optimize these prediction models in very preterm neonates and quantify the ability of metabolites to act as strong, robust and potentially longitudinal biomarkers of neonatal illness severity. We hypothesize that metabolic biomarkers can be used to accurately predict the risk of a composite outcome in very preterm neonates that includes neonatal morbidity and in-hospital mortality. The objectives of our study are: Aim 1: Develop and externally validate metabolic models for predicting neonatal morbidity in very preterm newborns; and Aim 2: Evaluate dynamic metabolic models for predicting neonatal morbidity at multiple time points within the first week of life. The proposed work will examine metabolic predictors of neonatal morbidity and mortality in a retrospective sample of approximately 8,500 very preterm births from California and 1,500 very preterm births from Iowa. Furthermore, we will evaluate the ability of metabolites to predict neonatal morbidity and mortality at four critical time points within the first week of life in a prospective sample of 500 very preterm newborns receiving care in the NICU at UCSF Benioff Children's Hospital (UCSF-BCH) and the University of Iowa Stead Family Children's Hospital (UI-SFCH). Understanding the relationship between specific metabolites and neonatal morbidity will lead to the long-term goal of improved diagnostics, more effective therapeutic agents, and a precision approach to clinical management of the very preterm neonate.